Long-term outcome in patients with rheumatoid arthritis (RA) is highly dependent upon aggressive pharmacological control of inflammation early in the disease course. Despite the importance of selecting the optimal medication soon after disease onset, there is no clinical or biomarker predictor of drug treatment response. A genetic biomarker would be particularly useful for drugs that block the inflammatory cytokine TNF-alpha (TNF), as these drugs are first-line biological disease modifying anti-rheumatic drugs DMARDs, yet induce remission in only ~30% of patients. In this application, our central hypothesis is that common genetic variants of modest effect size predict response to anti-TNF therapy. To test this hypothesis, we propose to expand upon our established multi-center collaboration and available GWAS data to develop (i) new statistical methods for conducting GWAS (estimating variance explained by common single nucleotide polymorphisms, SNPs), (ii) new informatics methods for defining treatment response in the EMR (which will allow us to collect many more samples for GWAS), and (iii) a novel framework for testing mechanism directly in human immune cells. Aim 1: Analyze GWAS data on ~1,200 RA patients to search for common variants that predict response to anti-TNF therapy. Aim 2: Use electronic medical records (EMR) at Partners HealthCare, Vanderbilt and Northwestern to define treatment response, and conduct a GWAS on ~1,200 additional RA patients treated with anti-TNF therapy. Aim 3: Test mechanism of action of alleles that predict treatment response to anti-TNF therapy in human immune cells. PUBLIC HEALTH RELEVANCE: A long-term goal of understanding the genetic basis treatment response in patients with rheumatoid arthritis (RA) is to improve care of patients with this common and debilitating disease. In theory, identifying specific pieces of DNA (alleles) that predict treatment response should aid in targeting therapy to the right individuals early in the course of disease before bone destruction occurs.